Definition: what is Schema.org?
Schema.org was initiated in 2011 by Google, Microsoft, Yahoo and Yandex and defines types and properties for semantically annotating web content: a page explicitly tells the machine that it describes a person, an organisation, an article or a product — including attributes and relationships. Search engines use the markup for entity recognition and result features; retrieval systems benefit from the unambiguous mapping of author, date and content type.
Core types for GEO
- Person, Organization: the entity foundation — with
@id,sameAslinks (Wikidata, LinkedIn) and consistent attributes. - Article, BlogPosting, TechArticle: publication metadata with author reference and ISO 8601 dates.
- Dataset: studies and open data — including
distributionfor CSV/JSON downloads. - FAQPage, HowTo: still useful as semantic declarations — but no longer a SERP rich result since May 2026; a citation effect in AI Overviews is not proven.
- DefinedTerm, DefinedTermSet: glossaries and terminology collections.
- ProfessionalService, LocalBusiness: service and location contexts.
Implementation: JSON-LD and the @id graph
The most important implementation rule: JSON-LD over microdata/RDFa. JSON-LD is separate from the DOM, easier to maintain and explicitly recommended by Google; inline microdata is considered legacy. The second rule: connect entities via stable @id URIs instead of duplicating facts — one canonical person entity that every page references via @id keeps the graph consistent and contradiction-free.
The most common audit errors: missing @id (entity cannot be anchored), invalid date formats (not ISO 8601), circular or bloated sameAs lists, and schema types that do not match the visible content. The ground rule since the spam systems of recent years: markup must mirror visible content — declarative markup without a visible counterpart is a risk, not a lever.
What schema does for AI visibility — and what it does not
A realistic framing: schema markup anchors entities machine-readably, clarifies authorship and publication time, and makes it easier for retrieval systems to map content to sources. It is not a direct ranking or citation lever: no system documents that markup alone produces citations. Schema belongs to the infrastructure category — necessary precision work whose absence hurts, but whose presence does not carry results on its own.
Related terms
Schema.org is the tool of entity anchoring and the bridge to the Knowledge Graph. Publication markup supports E-E-A-T signals. Implementation details: schema implementation.
Schema is infrastructure, not a magic wand
Structured data anchors entities, authorship and publication time machine-readably. That is necessary precision work — but markup without consistent content and sources produces neither rankings nor citations.
FAQ on Schema.org
Which format is recommended for schema markup? ▾
JSON-LD. It is separate from the HTML DOM, easier to maintain and explicitly recommended by Google. Microdata and RDFa still work but are considered legacy formats.
Is FAQPage markup still worth it? ▾
Not as a rich result in the Google SERP — that display has been discontinued (for most sites in 2023, remaining cases until May 2026). It remains useful as a semantic declaration of question-answer structures; a proven citation effect in AI Overviews does not exist. Important: only mark up FAQ content that is visible.
What is an @id graph? ▾
A schema setup in which every entity carries a stable URI (@id) and pages reference the canonical entity instead of duplicating facts. It keeps person, organisation and content data consistent across the whole domain — the precondition for reliable entity recognition.
Does schema markup directly help with LLM citations? ▾
Not provably in a direct way. Markup improves the machine-readable clarity of author, date and entity — which makes attribution easier for retrieval systems. Citations, however, come from source authority, passage quality and accessibility; schema is the infrastructure underneath.